1
Q
Define what evidence-based medicine is
A
“The process of systematically reviewing, appraising and
using clinical research findings to aid the delivery of
optimum clinical care to patients”
2
Q
What is evidence-based medicine based on?
A
-Clinical research (gathering evidence)
-Evaluating clinical research (judging evidence)
-Informing clinical decisions (using evidence)
3
Q
What does the evidence need to be for evidence-based medicine?
A
Reliable and relevant
4
Q
What is epidemiology?
A
The science of gathering evidence to
answer :
-What causes a disease?
-How can we best treat a disease?
5
Q
What type of studies are used to identify what causes a disease?
A
Observational studies
6
Q
What type of studies are used to best treat a disease?
A
Intervention studies
7
Q
What do all epidemiological studies need to be based on?
A
A well-defined population
8
Q
Give a reason as to why we would take a sample from the well-defined population rather than just using the whole population?
A
Using the whole population would be quite expensive
9
Q
What does the sample need to be in relation to the population?
A
Sample needs to be representative of the population
10
Q
In an epidemiological study, if you have the disease, what are you defined as?
A
A case
11
Q
In an epidemiological study, if you do not have the disease, what are you defined as?
A
Normal
12
Q
What are the 3 ways that cases can be defined?
A
-Statistical
-Clinical
-Operational
13
Q
What is the statistical definition of a case?
A
An individual who meets a predefined set of diagnostic or classification criteria for a particular disease or condition
14
Q
What is the clinical definition of a case?
A
An individual who is identified as having a disease based on a specific set of signs, symptoms and clinical findings
15
Q
What is the operational definition of a case?
A
A precise, practical set of criteria used in a study to decide whether an individual will be counted as a case
You decide a convenient arbitrary location and set the “threshold”
16
Q
A study is done to investigate lung cancer
What would be the cases?
What would be the normals?
What disease definition would this be?
A
Cases - people that had lung cancer
Normals - people picked from hospitals that did not have lung cancer
CLINICAL disease definition
17
Q
Do we always need to define a disease?
A
Sometimes the measurement is enough
e.g.
1) Instead of “low” and “high” myopia, just measure
refractive error
2) Instead of “ocular hypertension” just measure the IOP.
18
Q
What are the 2 types of study designs?
A
-Observational studies/medical surveys
-Intervention studies/experimental studies
19
Q
What do observational studies/medical surveys do?
A
-Measure level of disease or condition
-Identify factors regarded
as causal in the development of a disease (risk factors)
20
Q
What do intervention studies/experimental studies test?
A
Test a new treatment to see if it is effective
21
Q
What time period do cross-sectional studies look at health at?
A
Looks at health at a specific moment in time (a “snapshot”
view)
In the present
22
Q
Is there follow-up or tracking over time in a cross-sectional study?
A
No
23
Q
What thing are cross-sectional studies suited to study?
A
Prevalence
24
Q
Define point prevalence
A
The proportion of
persons in a population who have the disease in question at a
particular point in time
25
What can cross-sectional studies find?
Associations, say between a risk factor and a disease
26
Give the advantages of cross-sectional studies
-Quick
-Cost effective
27
What can cross-sectional studies not establish?
Causality
Cannot prove that a risk factor causes a disease
28
Can a cross-sectional study tell us which came first (disease or risk factor)?
No
29
True or false
Cross-sectional studies can tell us how a disease progresses over time and whether interventions can reduce its occurrence
False
Cross-sectional studies cannot do this
30
Define prevalence
Fraction of the study population who are cases at a specific moment (or over a short period of time)
31
How do you calculate prevalence?
Number with disease / Total in survey
32
What is crude prevalence?
Number of cases / Population size
33
What is specific prevalence?
Prevalence in a subpopulation (e.g. race, gender, ethnicity)
34
How should the figure 29 (25.1 ... 35.3) be read as a confidence interval?
-The prevalence in the sample is exactly 29%
-There is a 95% chance that the prevalence in the population could be anywhere in the range 25.1% to 35.3%
-However, it’s more likely to be close to 29% than far from it
35
What is a prevalence ratio?
The ratio of two prevalences
36
What does a prevalence ratio tell us?
Says how much more/less common a disease is in one group than another
37
What is a prevalence ratio is measure of?
Association (and maybe a cause)
38
What does a prevalence ratio that is different from 1 indicate?
An association between a risk factor and a disease
39
What does a prevalence ratio near 1 indicate?
Not an association between a risk factor and a disease
40
When a disease is rare, what will the odds ratio and prevalence ratio be?
Odds ratio and the
prevalence ratio will be similar
41
What is an odds ratio a measure of?
Association
42
What numerical values does an odds ratio range from?
Zero to infinity
43
What will the odds ratio be if there is no association between a risk factor and a disease?
Odds ratio is close to 1
44
When there is no association between a risk factor and a disease, why will the odds ratio be close to 1 but not = 1?
Due to things like sampling errors
45
What will the odds ratio be if there is a positive association between a risk factor and a disease?
Odds ratio > 1
46
What does a positive association between a risk factor and a disease mean?
Means that the risk factor increases the chance of having the disease
47
What will the odds ratio be if there is a negative association between a risk factor and a disease?
Odds ratio < 1
48
What does a negative association between a risk factor and a disease mean?
Means that the risk factor decreases the chance of having the disease
49
What is the interpretation of an odds ratio between 1 - 1.45 with positive association?
Small or no association
50
What is the interpretation of an odds ratio of around 3.5 with positive association?
Medium association
51
What is the interpretation of an odds ratio of around 10 with positive association?
Large/strong association
52
What is the interpretation of an odds ratio between 0.69 - 1 with negative association?
Small or no association
53
What is the interpretation of an odds ratio of around 0.3 with negative association?
Medium association
54
What is the interpretation of an odds ratio of around 0.1 with negative association?
Large/strong association
55
What is the criteria for evaluating the causal significance of an association?
Consistency, strength, specificity, temporality and coherence of an association
56
What do we mean by the consistency of an association?
Different methods of study in different populations yield the same conclusions
57
What do we mean by the strength of an association?
The effect of the risk factor is large and there is a dose response effect
(risk of lung cancer increases with the number of cigarettes smoked daily)
58
What do we mean by the specificity of an association?
E.g. Most (but not all) people with lung cancer are smokers
59
What do we mean by the temporality of an association?
Risk factor must be present before disease occurred
(smoking started before lung cancer was diagnosed)
60
What do we mean by the coherence of an association?
Should not be at odd with the known facts
(dose response effect for cigarette smoking and the known carcinogenic effects of cigarette smoke)
61
What are cohort studies also known as?
Longitudinal studies
62
What do cohort studies do?
-Follow a group of people over time
-Compare with surveys which exist at a single point in time
63
What time period do cohort studies look at?
Starts in the present and goes into the future
64
What is a cohort?
Group of people with a common characteristic
65
In a cohort study, when is data collected?
At multiple points in time
66
What can be monitored in a cohort study as a result of data being collected at multiple points in time?
Disease progression can be monitored as there is regular follow up
67
What is able to be determined in a cohort study as a result of data being collected at multiple points in time?
Causality
68
What is a cohort study suited for evaluating?
Long-term outcomes
69
Give the disadvantages of a cohort study
-Time-consuming (mulitple points)
-Expensive(repeated tests)
-Risk of Px's dropping out over time (Px's may lose interest or elderly Px's may pass away)
70
If I was really interested in glaucoma or cataract onset, is a 30-35 age group a good idea?
No as you will have to wait many years for Px's to present glaucoma or cataract symptoms
71
If I was really interested in myopia onset, is a 30-35 age group a good idea?
-No as many of these people will already be myopic
-Need to start younger
-Unless you are looking for special cases
72
You’re studying the development of diabetic retinopathy using a cohort trial
Do you include people with retinopathy in your cohort?
No as you are studying the development
(you don't want people to already have retinopathy)
73
How do you calculate prevalence for a cohort study?
Number of cases / Number in study
74
Formula to calculate the incidence rate for a cohort study
Number of new cases in 1 year / Number not already a case at the start of the year
EACH YEAR, YOU NEED TO SUBTRACT THE TOTAL NUMBER OF PEOPLE THAT HAVE THE DISEASE FROM PREVIOUS YEARS
75
What is a patient year?
1 patient followed for 1 year
Could be 1 patient followed for 1 year or 2 patients followed for 0.5 years etc
76
How many patient years would it be if 2 Px's were followed for 1 year?
2 patient years
77
How many patient years would it be if 50 Px's were followed for 3 years?
150 patient years
78
How do you calculate patient years?
Multiply the number of patient by the number of years
Number of patients x Number of years
79
A cohort study followed 300 patients for 2 years and found 15 new cases
What is the incidence in patient-years?
Patient years = Number of patients x Number of years
Patient years = 300 x 2 = 600
Incidence = 15/600
80
What must the original cohort include?
The original cohort must include people :
-Who have been exposed to the risk factor
-Who have not been exposed
-Who might get the disease
81
82
What is a case-control study?
Study that is used to investigate possible causes of one disease
83
What is the goal of a case-control study?
Goal is to identify risk factors that might be associated with a disease, or a condition
84
What are the groups of people that a case-control study compares?
Cases - set of people who have the disease being investigated
Controls - set of people who do not have the disease (but may have other diseases)
85
A study was completed to investigate the socioeconomic factors (risks/deprivation) that are associated with late presentation of glaucoma
-Participants were 220 people newly diagnosed with glaucoma
-They were split into a group with advanced glaucoma (110) and a group with early glaucoma (110)
-Those with advanced glaucoma are the cases
They presented late (i.e. They put up with problems for a long time before going to doctor/optometrist)
Why can you not calculate prevalence for this study?
What can you calculate instead?
Cannot calculate prevalence as you have selected the total number of participants (we don’t know the denominator)
Can calculate the Odds Ratio instead
86
What are 2 measures of association that we can calculate?
-Prevalence
-Odds ratio
87
When does confounding occur?
-The true cause of the disease is not recorded
-The true cause is associated with another factor that is recorded
88
How does confounding affect a study?
What does it make the results seem like?
Looks like the other factor is a cause of the disease
89
What process reduces confounding?
Matching
90
How does the process of matching work?
You can match cases and controls on any factor you think might be a confound (e.g. age, sex, weight etc)
91
What can you rule out once a factor has been matched?
You’ve artificially ensured that there is little or no difference between cases and controls
-This means you can no longer tell if the factor is a cause of the disease (as it is no longer part of the study)
92
Give the advantages of Case-Control studies
-Allow researchers to study conditions that are less common or take years to develop (advantage over Cohort Studies)
-Require less time and money (as they are retrospective studies)
-Can look for associations between several risk factors and a single disease, within a single study
93
Give the disadvantages of Case-Control studies
-Selection bias
-Recall bias
-Confounding
-Temporal ambiguity
94
What do we mean by recall bias?
Px’s who have the disease are more likely to accurately record their past compared to Px’s that do not have the disease
95
What do we mean by temporal ambiguity?
Don’t know whether the disease or the risk factors came first when looking at a particular point in time
96
What is a clinical trial?
Test of whether a treatment or some other intervention works
97
What are we interested in when carrying out a clinical trial?
-Safety of the treatment/other intervention
-Efficacy in curing a condition
-Side-effects
98
What type of test does a clinical trial need to be?
FAIR TEST
99
What are the types of clinical trials?
-Observational trials
-Intervention trials
-Cohort studies
-Case-control studies
100
What is the role of the optometrist on clinical trials?
-Recruitment and screening
Identify potential candidates for the trial by screening Px's
-Px care
Provide ongoing care
Monitor Px's – ensure they follow protocol, etc
Manage adverse effects and complications
-Data collection
Collect data on VA, IOP, sxs of dry eyes
-Interpreting results
Understand and apply clinical results to practice
101
Give the common challenges with clinical trials
-Recruitment difficulties
Particularly challenging when the study involves rare diseases or conditions (e.g. keratoconus or retinitis pigmentosa)
-Compliance
Ensuring participants adhere to protocol can be difficult
-Funding
Requires substantial financial resource
102
What is regression to the mean?
Describes how unusually extreme results tend to move closer to the average/mean the next time they’re measured
If something is very high or very low compared to normal, the next outcome will likely be less extreme - just by chance
Extreme results tend to be followed by more average ones, due to randomness
103
What group does a fair test need, other than a treatment group?
Control group
104
Does the control group receive the treatment?
No
105
How is the effectiveness of the treatment judged?
The effectiveness of the treatment is judged by comparing the treatment group with the control group
106
What is a placebo?
A placebo is a medically useless treatment (e.g. a pill with no active substance) that nonetheless has a positive effect on a person’s health, because the person believes it should work
107
What is a nocebo effect?
When the neutral treatment makes people feel worse – may occur when reporting side-effects
108
What is the effect in the treatment group?
Effect in treatment group = tablet effect + placebo effect
109
What is the effect in the control group?
Effect in control group = 0
110
Difference treatment – control = tablet effect + placebo effect
Difference treatment – control = tablet effect + placebo effect
111
What type of tablet does the treatment group get, tablet with active ingredient or inactive ingredient?
Tablet with active ingredient
112
What type of tablet does the control group get, tablet with active ingredient or inactive ingredient?
Tablet without active ingredient
113
If people dropout from the trial, should they still be counted and why?
YES as they may have dropped out because the treatment was ineffective or they may have dropped out because of another reason
114
What features must a clinical trial have?
1) Treatment & control groups
2) Treatment group gets the treatment being trialled
3) Control group receives best available treatment or (less good) a placebo treatment
4) Px's allocated to the groups by matching and/or randomisation
5) Px's unaware which group they are in
6) Assessors unaware which group they are in
7) Dropouts are kept in the data
115
Why are Randomised Controlled Trials (RCT) considered as the gold standard?
-Bias is minimised
Matching reduces confounding variables
Randomisation reduces confounding variables
Blinding removes bias introduced by both Px's and researchers
-Highest quality of evidence
Provide highest level of evidence for efficacy and safety of an intervention
-Can provide strong evidence of causality
Can isolate the effects of the intervention, permitting stronger conclusion about cause and effect
-Generalisability
When properly conducted the conclusions can be generalised to the broader population
116
What are the 3 different kinds of outcome measure?
-Continuous (e.g. IOP, VA, refractive error)
-Discrete (e.g. Lens Opacities Classification System III (LOCS III) Cataract grade)
-Discretised (happens when a continuous measure is broken into groups - high myopia vs low myopia)
OP 1
flashcards
Decks in class (13)
# Cards
-
History of Optometry58
-
GOC/COO and Opticians Act38
-
Introduction to Professionalism/Case History (Hx)27
-
NHS, GOS and Consent71
-
Contract Law and Law of Tort51
-
Ethics In Optometry, Equality Act, Mental Capacity Act, Power Of Attorney, Professional Boundaries42
-
Business Legislation/Other Legal Aspects To Consider In Practice114
-
Contact Lens Teach (Lab)26
-
Screening, Sensitivity, Specificity/Probability and Risk51
-
Statistics/Regression and Correlation/Error Analysis153
-
Sources of Information for Optometry34
-
Epidemiology116
-
Abbreviations50
